Factors Associated with Enrollment in Cancer Genetics Research

A recent focus in cancer research has been to increase overall participation rates and to identify the predictors of enrollment in clinical trials. Investigators have consistently reported low accrual rates of cancer patients in oncology clinical trials (1-4); the overall participation rate of all newly diagnosed adult cancer patients in clinical trials is estimated to be as low as 2% to 5% in the United States (5-9). Although these figures do not reflect the enrollment rates for the smaller group of individuals who were eligible for and offered trial participation, it is clear that there are barriers to participation in cancer clinical research that need to be addressed.

Previous studies have also shown that racial and ethnic minorities, as well as women and the elderly, are significantly less likely to participate in cancer clinical trials than whites, males, and patients <65 years (2, 10-16). Despite NIH federal requirements encouraging increased enrollment of women and minority groups in all clinical cancer treatment protocols, such groups continue to be widely underrepresented in these trials (9, 11, 12, 16, 17).

Although there have been substantial efforts to identify factors associated with patient enrollment in oncology clinical trials, little attention has focused on the comparable topic of subject participation in cancer genetics research. In cancer genetics research, it is important to make the distinction between protocols that incorporate genetic testing and counseling for at-risk individuals and epidemiologic protocols that look for genetic markers in a larger population of individuals. When deciding to enroll in these types of epidemiologic protocols, individuals must consider the implications of donation, storage, and ownership of biological samples, collection of family history data, as well as privacy and confidentiality concerns related to informed consent and documentation of genetic risk information (18-22). Studies examining public attitudes toward the donation and storage of blood samples for future genetic analysis indicate that most individuals have favorable attitudes toward these research components (18, 23). However, a 1998 population-based survey found that individuals of white race were more likely to express positive attitudes toward participation in genetics research in contrast to participants of other racial and ethnic groups (23). Another national survey of individuals from the general population found that non-Hispanic, black individuals and females were significantly less accepting of study components associated with genetics (18). Most research on the participation of individuals in cancer genetics epidemiologic protocols has been limited to studies of hereditary breast cancer, which have found consistently low representation of racial and ethnic minorities in hereditary breast cancer registries (24, 25). To date, participation rates in genetics epidemiologic research have not been examined for other cancer diagnoses.

We conducted a retrospective analysis of subject enrollment in an ongoing R01 NIH-funded study examining the role of MSH6 as a factor in genetic predisposition for colorectal cancer. The primary goals of this analysis were to determine overall participation rates for this particular cancer genetics epidemiologic study as well as to identify the clinical and demographic factors that were associated with enrollment. A secondary objective was to determine whether there were differences among enrolled subjects in providing consent for the various components associated with cancer genetics epidemiologic research.

Subjects with colorectal cancer who presented to the Gastrointestinal Oncology clinics at Dana-Farber Cancer Institute and Massachusetts General Hospital (Boston, MA) between June 2001 and August 2004 were approached to participate in a genetic research study examining the role of certain genes as predisposition factors for colorectal cancer. During the initial 2 years of recruitment, only patients with a family history of gastrointestinal cancer in a first- or second-degree relative were approached for enrollment. During the last year of the study, all individuals with a diagnosis of colorectal cancer were recruited for enrollment, regardless of their family history of cancer. The change in recruitment strategy reflects an expansion of this genetic banking protocol to include individuals with sporadic cases of colorectal cancer as well as control cases of individuals who had not been diagnosed with cancer. The potential difference in enrollment rates due to this change in recruitment is controlled for in the multivariate analysis.

The main component of study participation involved providing a one-time blood sample for genetic analysis for mutations in the MSH6 gene, which is involved in DNA mismatch repair mechanisms. The necessary requirements for study enrollment included signing a written informed consent as well as agreeing to provide the blood sample. Individuals who either fulfilled classic Amsterdam criteria for hereditary nonpolyposis colorectal cancer (26) or were <18 years were ineligible to participate. Additionally, individuals who did not speak fluent English and for whom translation services were not available (n = 20) or who seemed to be acutely ill (n = 20) were not approached for study enrollment. Overall, 1,111 subjects with a diagnosis of colorectal cancer were asked to participate in this study. All study procedures and documents were approved by the Dana-Farber/Harvard Cancer Center Institutional Review Board.

Eligible subjects were identified through the daily review of the new patient gastrointestinal clinical schedules at both recruitment sites. From these schedules, a total of 1,111 individuals diagnosed with colorectal cancer were identified and screened by a research coordinator for study entry. Over the course of the 3-year study period, there were three research coordinators who recruited subjects for study enrollment. The research coordinator approached every eligible patient to discuss enrollment, explaining that the main purpose of the study was to identify predisposition factors for colorectal cancer. Although a standardized script was not developed for obtaining informed consent, each research coordinator focused on key sections of the study consent form that outlined the risks and benefits of study participation to individuals considering enrollment. It was made clear to the potential study subject that individual genetic analysis results would not be disclosed; therefore participation in the study would not be of personal benefit to those who enroll. However, these individuals were informed that the overall study findings could potentially benefit individuals in the future with regards to using genetic testing to screen for an individual predisposition for developing colorectal cancer. During the 5 to 10 minutes of presentation of the study, subjects were asked to provide permission for study staff to obtain the following: (a) a one-time blood specimen for MSH6 mutation analysis, (b) a paraffin block of their tumor, (c) family history and data about risk factors for colorectal cancer, (d) medical records for review, and (e) permission to establish cell lines from the blood specimen. Individuals were also given the option to consent to be recontacted to consider the possibility of discussing the study with their first- and second-degree relatives and/or to be notified about future studies. The main requirement for study enrollment was that the subject agreed to provide the blood sample for genetic analysis and to sign a written consent. Consent for the remaining study procedures was not necessary for an individual to participate in the study and was left to the discretion of the study participant. Of note, the options to consent to the establishment of cell lines or to be recontacted in the future to discuss enrollment in additional studies were not included in the consent form during the 1st year of recruitment for this study. These two consent form questions were added during the 2nd year of the study due to a modification to this genetic banking protocol; therefore the response denominator for these consent questions differs. Total time required for study participation was ∼20 minutes. All 1,111 individuals approached for study enrollment were included in the present analysis.

Demographic information was obtained from a self-administered clinical form that is completed by all new patients with colorectal cancer before their initial clinical visit. This form required patients to provide a detailed personal medical history as well as family history data relating to history of cancer diagnoses. Family history was obtained by patient self-report and confirmed with the medical oncologist's dictated clinical note.

Descriptive statistics of the overall subject characteristics were generated. Categorical variables were then compared using χ² and Fisher's exact tests. Predictors that were found to be significant in univariate analyses were included in a standard multivariate logistic regression analysis, which was adjusted for year of recruitment and clinical location.

In addition, tests of interaction for the age and gender variables were done in the multivariate model and adjusted for other significant variables. All analyses were done using SPSS for Windows software (version 12.0; SPSS, Inc., Chicago, IL) and SAS statistical programs (version 8.0). All Ps were two sided, and significance was defined at P ≤ 0.05.

A total of 1,111 patients were approached for enrollment (Table 1). Sixty-six percent of patients were age ≤65 years and 52% of patients were male. The vast majority of patients were white (95.8%); 2.1% were black, 1.2% were nonwhite Hispanic, 0.9% were Asian, and ethnicity was unknown for 0.9%. Nearly one fourth of patients presenting with colorectal cancer reported having a first-degree relative with colorectal cancer (23.2%), and 82.3% of patients had children. The majority (92.9%) of individuals were presenting to clinic for an initial visit after being diagnosed with colorectal cancer. Approximately two thirds of patients had private health insurance (67.1%), whereas 27.3% of patients had Medicare, 3.5% had Medicaid for their primary coverage, and 2.1% were uninsured.

Of the 1,111 eligible subjects, 696 (62.6%) enrolled in the study (Table 1). In univariate analyses (Table 1), patients who enrolled were more likely to be ≤65 years of age [odds ratio (OR), 1.43; 95% confidence interval (95% CI), 1.11-1.85], male (OR, 1.44; 95% CI, 1.13-1.84), and white (OR, 2.06; 95% CI, 1.14-3.74). In addition, patients with a first-degree relative with colorectal cancer were more likely to participate (OR, 1.36; 95% CI, 1.01-1.83). Visit type, insurance type, or whether patients had children was not significantly associated with enrollment (Table 1). The results of multivariate analysis are displayed in Table 2 and similarly show that older individuals, females, racial/ethnic minorities, and those without a first-degree relative were significantly less likely to enroll. The multivariate analysis was adjusted for year of recruitment as well as recruitment location, and neither of these factors had a significant effect on the study enrollment rates.

We also explored age and gender interactions that may influence enrollment, based on previous reports that suggest that these variables may play a role in study participation. The interaction model for age and gender showed a significant interaction (P = 0.035). In both age groups, men are more likely to enroll than women, but the gender effect is stronger among individuals ≥65 years (OR, 2.32; 95% CI, 1.50-3.59) than among the younger individuals (OR, 1.29; 95% CI, 0.94-1.77). The small number of nonwhite patients in this cohort limited our ability to explore interactions between ethnicity, age, and sex (Table 3).

Table 4 displays the percentage of enrolled subjects who declined the study components outlined in each of the seven consent form questions. Overall, 88% of enrolled subjects agreed to participate in all study procedures. None of the enrolled subjects declined to provide a blood specimen for genetic analysis or to provide permission for researchers to access medical records and pathology reports related to the subject's cancer diagnosis. A small number of individuals declined to give permission for researchers to establish cell lines from their blood sample (1.6%), and one subject (0.1%) declined to give permission for researchers to access radiological reports related to the subject's cancer diagnosis. With regards to recontacting study participants in the future, a small percentage of participants declined to give permission for study staff to recontact them to discuss enrollment in additional research studies (3.6%) or to discuss the possibility of enrolling additional family members in the study (3.4%), and 15 individuals denied permission for additional contact to ask for additional blood or tissue samples (2.9%; Table 4).

The overall enrollment rate in our study of 63% of eligible patients seems to be higher than those participation rates reported for many cancer treatment trials. Racial/ethnic minorities, individuals without a first-degree relative with colorectal cancer, as well as females >65 years were among those least likely to enroll. Interestingly, disparities in study accrual by age seemed to be a function of gender. In this study, men were more likely to enroll in the study than women, and this gender effect was found to be stronger in older individuals than in those individuals ages ≤65 years. In addition, once subjects decided to enroll, most agreed to consent to all genetic study components; only a small percentage of participants declined consent to certain study procedures.

A review of enrollment rates in cancer treatment trials reveals that the overall participation of eligible patients may range from 30% to 69% (6, 7, 27) of those recruited. Participation in cancer genetics epidemiologic research is less well documented, and previous reports indicate that the response rate for these types of studies is variable, depending on the study setting and patient population (24). Despite the 63% participation rate for our study, it is difficult to compare accrual rates for cancer clinical research studies, given that each protocol has its own target population, eligibility criteria, and study intervention requirements.

Specific concerns for individuals considering enrollment in epidemiologic genetics research may relate to the confidentiality of genetic information, disclosure of test results, and fear of medical and insurance discrimination (18-20, 22). Published reports have shown that patients consider health insurance coverage to be an important factor when considering enrollment in oncology treatment trials (6, 7, 11, 27). Although we expected that potential subjects might have concerns about whether genetic information would be shared with insurance companies, we did not find that the type of health insurance influenced a subject's decision to enroll.

Hereditary cancer research involves the review not only of individual medical records but also of family history data, which could raise fears about the privacy of personal information for study participants and their family members. The possibility of recontacting a subject to confirm the family history of cancer or to discuss study enrollment of additional family members could be viewed as bothersome for subjects and may deter individuals from enrolling in these types of studies (21, 22). In spite of these issues, previous studies in human genetics have revealed that subjects are not overly concerned about the confidentiality of their medical records information or donating samples; in fact, most individuals have stated that they do not view these interventions as risky or distressing (18, 23, 28). Our results support these data, as the majority of enrolled individuals were comfortable with donating a blood sample as well as permitting the review of tissue samples and medical records for study analysis. In addition, only a small percentage of subjects were unwilling to be recontacted to provide additional samples (2.9%) or to discuss the possibility of approaching other family members for study enrollment (3.4%).

On deciding to participate in the study, the vast majority of individuals (88%) tended to agree to consent to all study procedures, whether they were associated with genetic components or not. It seems that the main barrier to study participation was providing a blood sample for genetic analysis, as this was the only official requirement for study enrollment in addition to signing a consent form. Future investigations are warranted to explore which aspects of participation of cancer genetics research present the greatest concerns for those individuals who decline enrollment in these studies. In addition, previous reports have documented that patients may not understand or remember the study consent process, which raises questions about whether patients recognize the requirements and risks of their involvement in research (10, 29-32). Despite the efforts to obtain informed consent, our results raise the concern that patients may not appreciate the nuances among different study components (e.g., the differences between DNA analysis and the creation of cell lines). Given the potential legal, medical, and ethical implications associated with cancer genetics epidemiologic research, it is important to further explore the study consent process for these types of protocols (18, 20-22, 25, 31).

The low accrual of minority subjects in our study is comparable with previously published studies where racial/ethnic minorities are consistently less likely to enroll in research studies compared with white subjects (1, 12, 18, 24). A potential explanation for the low enrollment of minorities in our study is that translator services were not always available at the patient visit, making it difficult for research staff to have a well-informed discussion about the study with the potential subject. Given the low number of minority subjects in the recruited study population, we do not have enough accurate information about an individual's primary language to assess the degree to which language served as a barrier to study participation for different racial/ethnic minority groups. In several reports, certain racial and ethnic minorities have expressed a greater distrust of medical research compared with patients of white, non-Hispanic backgrounds (15-17, 33, 34). Moorman et al. (25) speculated that nontraditional family structures among some minority patients may deter enrollment in cancer genetics research because of the requirement of a well-documented account of the family history of cancer.

We had anticipated a lower participation rate for individuals ages ≥65 years as well as females based on previous reports of study enrollment in cancer clinical trials (11-14). Our analysis shows that there is an important interaction between age and sex, with women ages ≥65 years accounting for the group with the lowest participation rates. Murthy et al. (16) reported similar results in an analysis of enrollment fractions for colorectal cancer trials, where men >65 years were more likely to enroll in available treatment trials than women of the same age group.

There are several limitations to this study. First, the overall number of ethnic/racial minorities in the eligible study population is low, making it difficult to interpret the noted disparity in study enrollment between white and nonwhite groups. The lack of minority patients in our study population creates a selection bias, in which we were only able to analyze data of patients referred to the gastrointestinal oncology clinics at our centers, excluding access to information on patients that could be presenting to other Boston-area clinics for treatment options. Another limitation of this analysis is that we only had access to certain clinical and demographic data from the patient clinical form; we were therefore unable to measure all potential factors that could influence an individual's understanding and perception of hereditary cancer syndromes, including education level and socioeconomic status. The lack of access to particular medical records information, such as stage of colorectal cancer diagnosis for study subjects, could be an important factor in an individual's decision to enroll in a research study. It is likely that those patients who were too ill to be approached for study enrollment had progressed to a later stage of disease; therefore our study population may be biased toward selecting for individuals with early-stage colon cancer who were more likely to enroll in the study. An additional limitation is that, under this protocol, we did not have the option of asking patients about their attitudes and beliefs toward human genetics research, information that would be helpful in explaining reasons why patients choose to enroll or to decline participation in these types of studies. We were also unable to analyze factors related to the consent process, such as the average time spent discussing the study with the patient as well as the influence that family members play in an individual's decision to enroll.

Despite these limitations, our findings provide useful information on the largely unexplored topic of participation in cancer genetics research. This analysis is one of the first to look at the enrollment of subjects in a colorectal cancer genetics study, as previous reports of this type have focused on participation in hereditary breast cancer registries. An additional strength of our analysis was that we were able to examine demographic information on those individuals who declined study enrollment, providing us access to information not usually available for analyses of participation in clinical research protocols.

Future research to improve enrollment in cancer genetics research should examine patient attitudes toward research and knowledge of cancer genetics, primarily among underrepresented groups, such as minorities and elderly females. This information could help improve techniques and tailor recruitment strategies when presenting cancer genetics information to individuals of various age, gender, cultural, linguistic, and family backgrounds. To improve recruitment to genetics studies and to improve the quality of the informed consent process, it would be helpful to devise ways to communicate information about genes and heredity, as well as the meaning of terms, such as cell lines and DNA banking. In the clinical setting, strategies to improve accrual to colorectal cancer genetics research could include involving physicians or nurses during the study presentation and informed consent process to cover all of the questions or concerns raised by the patient. Overall, additional investigations are necessary to identify other factors that may influence enrollment in these types of clinical research studies to improve accrual to these studies, particularly among minority groups.